← Home

pov: you’re idly musing about getting a ref sheet in public

Experiment

On 2025-01-26, at 9:36am UTC, I made the following skeet on my Bluesky account:

Dammit I need a ref sheet of my human form, don’t I

On 2025-01-28, at 1:08am UTC, I made the same tweet on x.com.

Hypothesis

Nothing would happen.

Result

As of time of writing (2025-01-29, around 4:30am UTC), the posts collectively received 39 likes, 34 replies, and 16 follows, across 50 distinct accounts. (Multiple replies from the same account were collapsed. Even so, these numbers don’t match up with what the websites report, and I give up trying to make them match.) Five of the likes were from my friends and were excluded from further analysis, leaving n = 45 accounts (23 on Bluesky, 22 on Xitter — note also the substantial time gap between my posts on the two platforms, meaning that I reached about the same number of interacting accounts in half the time on Xitter, though to their credit they did hide maybe half the replies under the “Show additional replies, including those that may contain offensive content” interstitial). Almost all of these remaining accounts appeared to be responding to recommend I commission them for a ref sheet or start a conversation that might lead to that outcome.

I may update these numbers and observations if I get more replies, but I’ll probably become bored of this soon.

(Kinda) Quantitative observations

Note: All information was hand-labeled by me scrolling until I got tired, so data accuracy is not guaranteed.

Also note that, while most criteria below were intentionally chosen for affecting how much I’d want to commission an artist, not all of them are directly linked. In particular, many legitimate professional artists I’ve commissioned don’t post anything about their personal lives, and I completely respect that.

• All 45 accounts (100%) had some historical activity of responding to similar tweets to self-recommend or start a conversation, though I classified 6 as borderline accounts that sometimes made some replies that were unrelated to commissions, or were simply non sequiturs.
• 44 (98%) accounts displayed at least one example of their prior work, though 3 accounts only displayed one example and 2 accounts only displayed two. The last account with no examples of their prior work was an account with borderline self-recommendation activity.
  • 13 (29%) accounts had posted prior commissions but included no information about the commissioner, beyond vague descriptions like “client on Discord”. There were 3 borderline cases, and 1 account whose commissioners included an unexpectedly large number of suspended accounts.
  • One account claimed to have done a commission for “AlexJonesYT on Twitter”. Said account had their profile decked out as if they were in fact Alex Jones of Infowars fame, but was inactive enough that I suspect they were (also) an impostor.
• 36 (80%) accounts had no terms of service for commissioning. 2 remaining accounts had a ToS I subjectively decided was bad. Note that I visited some links from accounts’ bios to look for a ToS, but did not check anything requiring a login, e.g. joining Discord links.
• 35 (78%) accounts had no list of prices for commissions. 4 remaining accounts had lists I subjectively decided were bad.
• 29 (64%) accounts shared no personal information and made no personal posts. 15 accounts shared only personal information I judged to be “weirdly impersonal”, e.g. posting photos of generic objects, vaguely referencing that they were taking classes or recovering from a sickness with no further details. The 1 remaining account liked some specific video games and anime.
• 15 (33%) accounts had an inconsistent art style, in a way that suggested their art was not all the original work of one artist. I tried to be somewhat strict about this determination: in the most extreme cases, some accounts’ portfolios spanned a variety of traditional and digital illustrations, animations, VTuber rigs, cover art for books and movies, and pixel art, with no consistent style or themes whatsoever. Still, there were some fairly subjective determinations.

Profile pictures

• The majority of accounts (23, or 51%) had what I judged to be an anime profile pic, depicting a human. The vast majority were girls, though as it was hard to classify unambiguously in every case, I did not further break down these pictures by gender. (I also included among 23 what was, by the account’s own implied admission, a Studio Ghibli crop.)
• 6 (13%) accounts had a photograph (or possibly a photorealistic generation) of a human.
• 4 (9%) of accounts had a furry profile picture.
• 11 (24%) of accounts had a profile picture depicting a human in some other style. I would vaguely describe many of the styles as Western illustration, but don’t know more precise categories.
• 1 (2%) account had a profile picture depicting an inanimate object.

Subjective observations

I spent a decent amount of effort looking for hard evidence of one of these accounts directly plagiarizing another artist, but it was harder and I found less of it than I expected. I put a lot of images into Google Reverse Image Search and Tineye with no results, so I stopped doing that after maybe half the accounts in this analysis. However, I did find:

• one account displaying on their portfolio, among tons of other genres of art, a single piece of pixel art with an obvious name, “therickpixels”, written on it. therickpixels appears to be a legitimate pixel artist who had many other pieces with the same style and same pixelated signature, making this the most blatant case of plagiarism I found.
• one account vaguely implying they drew a cartoon that actually came from Helluva Boss.
• one account whose profile picture had, in tiny letters, the name PinkBites, who had in the past offered YCH commissions that just about perfectly matched the profile picture. Strictly speaking, I cannot rule out that the account legitimately commissioned PinkBites for one such YCH to use as their profile picture and merely failed to credit them, though of course that’s still not great.
• one account whose reply to me included an example dragon ref sheet. Not only did that account not understand the assignment, searching for the name on that ref sheet revealed that it was by a different artist, IcelectricSpyro, whom I had actually commissioned in the past!

There were many pieces of art that I thought could plausibly have been AI-generated and a few where I would guess were more likely than not to be such, but almost none that I would confidently describe as such. Among hundreds I examined, I only found two images I was almost 100% sure were AI-generated, a movie poster on one artist’s portfolio and an illustration of an anthro dog in a spacesuit on another artist’s account, both of which featured text with typically AI-generated glitches. In both cases, half the words were misspelled or had Frankenstein letters fused from multiple English letters. Of the art I didn’t think was AI-generated, there was a substantial fraction I might compare to the wikiHow house style.

Other observations:

• Many accounts were not shy about bemoaning how post replies were full of bots while replying to my post, or indeed in replies to other posts I saw on their accounts. One reply said, “Are you fr or is it just a bot bait>?”
• Many accounts were much more aggressive about watermarking their art than other artists I’ve actually commissioned and worked with.
• Many accounts accepted commissions through vgen.co, a commission website I hadn’t heard of.
• A lot of accounts made (or claimed to have made) streaming/vtuber assets, sometimes with bizarre juxtapositions. I laughed out loud at a piece with the text “GAMING SOUTH AFRICA”, a vector lion in the background, and a glowing Goku in the foreground. I’m told the glow comes from something called “Ultra Instinct”.

• Several accounts’ terms of service appeared to come from the same template, with long phrases repeated verbatim between them with only small changes to the numbers. For example, they said that turnaround time would vary…

  depending on my schedule and other factors such as my mental health, school, complexity of the work, and other personal issues that may occur.

• My favorite ToS section was a REFUNDS section that simply read:

  REFUNDS
  We can talk about it

• Several artists posted a line art version of their art alongside it, perhaps to imply that their art was truly hand-drawn; I was convinced in a few cases at first. However, in many (not all) cases, the line art could be observed to be the colored/shaded art after a black-and-white contrast filter.
• One artist displayed mobile Discord screenshots of them delivering commissioned art to happy clients, perhaps trying to demonstrate that the commissions genuinely happened, but it certainly had the opposite effect on me, raising my suspicion that they were all staged.
• One artist self-described as a wolf, but described and tagged the art with their profile picture on FurAffinity as “Dog (Other)”, even though other pieces of art on their FA page were tagged as “Wolf”.
• I got two friend requests on Discord (which isn’t linked to either of my social media accounts, but is easy to guess).
• This is now my most-liked post on Twitter and second-most-liked post on Bluesky, the latter only losing to me posting a commission I got. It’s a low bar, of course. But it does suggest an extremely funny way to artificially boost one’s follower counts and engagement for free. Who needs to pay for fake followers on the black market?

Conclusion

I am still looking for an artist to commission a human ref sheet from, but am at serious risk of deciding to just draw one myself.